The present invention relates to managing the provisioning of computing resources, and more particularly to determining optimal timeouts and optimal polling intervals for containers and virtual machines.
Cloud provisioning operations and transactions are complex and highly environment-dependent. The time required for a particular cloud provisioning operation may vary depending upon the context. For example, if an instance is a container (e.g., docker), the time required for a cloud to provision the instance is significantly less than the time required for the same cloud to provision a normal virtual machine instance. A timeout associated with a cloud provisioning operation or another operation is a period of time after which the operation is interrupted and/or an error condition is raised if the operation has not been successfully completed.
The monitoring of any operation or cloud object (e.g., application) includes regular checks (also known as polls) of the operation or cloud object which determine the operational status of the operation or cloud object. A polling interval is an amount of time between the start of two successive polls of an operation or cloud object. The polling interval is based on the temporal frequency of the polls of the operation or cloud object. The appropriate polling interval of a monitored operation or cloud object is related to the type of operation or cloud object being monitored, the environment of the operation or cloud object, and the performance conditions to which the monitored operation or cloud object is being subjected.
A cloud service may be unavailable in one or more clouds because a non-optimal timeout and/or a non-optimal polling interval is associated with an operation that is utilized by the cloud service.